What statistical test should I use?

#1
I am currently doing my final year project on the topic of depression among undergraduate students. I want to determine what are the factors that cause this problem. The statistical methods that I'm going to use is pearson's correlation of coefficient , chi-square test and multiple regression. For multiple regression , i am confuse whether I should choose a stepwise multiple regression , forward or backward? Also , if possible please tell me if the statistical methods i choose is correct or not.
 

Karabiner

TS Contributor
#2
Could you please describe your data first? The statistical methods depend on what data you have.
Which variables are included into the study, and how were they actually measured (e.g. how
exactely was depression measured). How did you recruit your sample and how large is the
sample size?

With kind regards

Karabiner
 
#3
Could you please describe your data first? The statistical methods depend on what data you have.
Which variables are included into the study, and how were they actually measured (e.g. how
exactely was depression measured). How did you recruit your sample and how large is the
sample size?

With kind regards

Karabiner
So i am going to collect my data using questionnaire which consists of two part. The first one is socio-demographical data and second one is the depression score determine by the BDI. The sample size had not defined yet but it will definitely be at least 200. I am using stratified sampling method . I already attach the questionnaire i'm going to use
 

Attachments

Karabiner

TS Contributor
#4
Your dependent variable can be assumed as interval scaled, I suppose.
You could consider multiple linear regression. Do not use stepwise (it
produces non-generalizable results and distorted p-values). Choose the
predictor variables for your model based on prior knowledge about
depression-associated factors, and on what would be interesting for
you to find out with your study.

With kind regards

Karabiner
 
#5
Your dependent variable can be assumed as interval scaled, I suppose.
You could consider multiple linear regression. Do not use stepwise (it
produces non-generalizable results and distorted p-values). Choose the
predictor variables for your model based on prior knowledge about
depression-associated factors, and on what would be interesting for
you to find out with your study.

With kind regards

Karabiner
should i include Cronbach alpha? for the reliability.......
 

Karabiner

TS Contributor
#6
alpha for what? For the BDI? That depends on what your instrcutors expect from you.
Usually I would assume that you do not have to perfom another such analysis on an
already established and well-documented instrument.

With kind regards

Karabiner